This is a subject which fascinates me more than many others, and my love of Asimov and other science-fiction authors is either the cause or the effect of this too. If you grew up reading and watching philosophical investigations into what is is to be human, conscious, creative through the lens of robotics then this whole explosion of A.I. art and writing is the discussion you’ve been waiting to have for real your entire life 😀
Asimov developed the notion of the positronic brain – a sort of hand-wavey bit of technology in order to have what would otherwise be a mechanistic bit of technology behave non-mechanistically and, therefore, circumvent having to explain what consciousness actually is. It is difficult to explain consciousness without, at some stage, having to introduce an and then some magic happens element regardless of whether it is literal woo or just magic by the back-door by simply invoking complexity. If it cannot be explained step-by-step, it may as well be magic.
Creativity is a similarly woolly term. If we could explain it mechanistically then it’s obvious that it could be replicated by an algorithm. If we cannot explain it as such, then it is clear we do not sufficiently understand it to be confident precisely when it is or isn’t occurring. We are really only left with a kind of gut-feeling – it cannot be creative because it is not complex enough, although we cannot state how much complexity is required. I suspect the answer is that no amount would be enough because, fundamentally, many people would like to believe that human beings are more than simply organic machines. That a computer could never be creative or, heaven forbid, conscious because human beings are special and some form of (for lack of a better term) divine spark is essential.
Whether or not A.I. art, as it exists currently, is any good is to me the least interesting discussion around all of this. I am, however, more broadly interested in the discussions around whether it is ethical or copyright-infringing because this, whether people like it or not, does start treading into this woolly topic of creativity. How, after all, do humans create art? If I want to draw a cat, I can draw on a lifetime of seeing cats or I can draw this specific cat which is in front of me. This is data-scraping albeit in a form where no copyright can be said to be infringed. But this data also includes every other work of art I have seen, or book which I have read and the subject of when inspiration becomes plagiarism is complex, hard to pin down precisely and, really, more convention than absolute. How unlike the dataset would the results have to be before people stopped objecting to A.I art?
My suspicion is that there would never be a point that this would happen, because the objection seems to be as much that human art exists in the dataset at all as that the results often resemble those sources. I’m not convinced that if the process of manipulating that data became advanced enough that it would pass a plagiarism test applied to an equivalent piece drawn by purely human hands that the objections would cease. My gut instinct is that this is more about not wanting to concede the magic of human creativity to a mechanistic process. We’ve seen many skills, tasks, and jobs replaced by robots more cheaply and efficiently over time but art, creativity, well all that was safe. I’m not sure that it is and, furthermore, I don’t think we’re all that far away from it. I accept that I probably have a significantly more generous view of when something might be said to be creative, or when something can be called art than many other people. But the thing is, that no-one is correct about these terms when they are so woolly. It all just comes down to acceptance, convention, and agreement rather than there being a rigid test that must be passed in order to declare, “this qualifies as art”.
Is typing in a prompt in some A.I. art or writing tool seeding a creative element, or is this programming? Or both? Programming languages have developed over time from machine code through to things which read more like English, or are represented by flow-charts. That we’d eventually be able to program by simply providing a set of simple English-language instructions, like Star Trek personnel might “program” the HoloDeck, seems inevitable and in no way ceases to be programming. Sure it’s programming of the most accessible kind which all but destroys the jobs of hard-core coders and, similarly, if I can just describe the scene in the style I want and get the art I want we’ve all but destroyed the jobs of artists too. The ability to create beautiful art being placed in the hands of people with no artistic skill what-so-ever is just accessible art. Why should the process of being able to create this stuff be exclusively in the hands of those blessed with the skill to do so? If we push back about this, why then praise game creation tools being accessible to those without deep knowledge of code? Would we object if A.I. was data-scraping GitHub to produce algorithms with the same gusto as we do art?
With my Wacom pad and the right software, I can simulate oil painting pretty convincingly digitally without any of the requirements of learning how to actually properly oil paint. I can access any colour I like from a convenient colour wheel without learning anything about how to mix pigments. I can mix in photographic elements into my paintings (which some would still consider cheating) or even directly trace it. I can augment my skills with any number of algorithmic filters. There are myriad digital painters who wouldn’t be able to paint for toffee on a canvas. Are we really going to draw a line somewhere and say, “this much, but absolutely no more automation” or does that line not inexorably move over time and where that line is for you is really just a function of how much automation you are used to due to what was available when you learned to draw?
I don’t have any answers to these questions. It’s all complicated and filled with, “well… it depends” type responses. I just feel that arguing about what is or isn’t creative, art, plagiarism, copyright-infringing, etc. is really just the dying wails of clear eventual obsoletion. Do not go gentle into that good night.
It is interesting to me, that in the Star Trek universe they still have chefs, people playing flutes and trombones, works of art and craftsmen even though literally all of that stuff has been made obsolete. I think that this is not far from the truth of what the future will hold – no matter how much of this stuff can be replicated faster, more efficiently, by algorithms and machines that there will still be value in the human craft. That we will attach greater value to something purely on the basis that we know it was made by human hands, in the same way that a guitar played by Elvis has more value attached to it than a guitar which was not even if those two guitars are otherwise identical.